Tech Times

Diffusion LLM Learns to Be Its Own Draft Model: NVIDIA Releases Tri-Mode Open Weights

Nemotron-Labs-Diffusion, NVIDIA’s new tri-mode language model, eliminates the separate draft model in speculative decoding: ...
GitHub

ARDY: Autoregressive Diffusion with Hybrid Representation for Interactive Human Motion Generation

ARDY is an autoregressive diffusion model designed for interactive motion generation, supporting online text prompting and flexible long-horizon kinematic constraints (root paths/waypoints, full-body ...
Tech Times

NVIDIA’s New LLM Decodes 6x More Tokens Without an Auxiliary Draft Model

NVIDIA Nemotron-Labs-Diffusion is a new tri-mode language model that eliminates the separate draft model in speculative ...

NVIDIA’s New AI Model Could Make ChatGPT-Style Responses Much Faster

Nvidia introduces Diffusion Nemotron, a 60-billion parameter two-tower AI model that uses parallel block processing to speed up text generation.
GitHub

diffusion_distillation_tutorial.review.json

"notes": "Solo subagent (low codex contention) wrote 1570-line draft with 3 [needs-verify] markers (Flow-OPD scope / rCM acronym / Rectified Diffusion brief mention)." "Updated ADD `L_distill` ...
IEEE

Diffusion Model-Based Visual Compensation Guidance and Visual Difference Analysis for No-Reference Image Quality Assessment

Abstract: Existing free-energy guided No-Reference Image Quality Assessment (NR-IQA) methods continue to face challenges in effectively restoring complexly distorted images. The features guiding the ...
IEEE

DLDiff: Image Detail-Guided Latent Diffusion Model for Low-Light Image Enhancement

Abstract: Low-light image enhancement is an essential task in image restoration. Inspired by the diffusion model, the related methods have achieved remarkable results in low-level visual tasks.